Laserfiche WebLink
<br />. <br /> <br />. <br /> <br />. <br /> <br />~, .. , <br /> <br />The State will provide the contractor with an Access Template that will be populated with the <br />Irrigated Acreage coverage attributes in the format required by the State. The contractor will <br />export the INFO tables in a .DBF format, which will in turn be imported into the Access <br />database using the State's Access template. <br /> <br />Deliverable to Contractor <br />Access template for data population. <br /> <br />Deliverable to State: <br />A task memorandum describing the process. <br />Files: An MS-Access database or databases (depending on template structure) <br />containing the attributes from the Irrigated Acreage coverage. (This includes <br />polygon/structure ID tie.) The Access template populated with the .PAT from the <br />Arclnfo polygon coverage and the additional structure information (2 tables with <br />the information for all divisions). <br /> <br />The .img file is part of the deliverable in Task 18. <br /> <br />Task 15. Metadata <br /> <br />Metadata similar to the metadata tables provided on Division 4 in the original RFP will be <br />produced for each final coverage delivered. <br /> <br />Deliverable to State: <br />Files: Metadata for each final coverage in HTML and metformat. <br /> <br />Task 16 - Supervised Classification - Guided Clustering <br /> <br />Processing Purpose: Classify crop types and irrigated acreage to achieve the accuracy goals <br />stated of the project. <br /> <br />Classification will utilize both supervised and unsupervised methods. In the supervised <br />process, field data are collected prior to classification and used as input to train the software to <br />recognize each different land cover category. The standard supervised methods be <br />augmented by guided clustering protocols utilized by Lillesand in the WISCLAND land cover <br />mapping project for the State of Wisconsin will be used. Guided clustering will improve the <br />ability to select spectrally pure signatures, will facilitate the editing of signatures and help to <br />remove non-crop and non-irrigated areas due to mis-registration in the dataset. <br /> <br />Guided clustering, like traditional supervised classification, is initiated with the image analyst <br />delineating a number of training areas to represent the various information classes to be <br />discriminated. Then, the pixels for each information class are independently sorted into a <br />series of spectrally homogeneous sub-classes using an unsupervised clustering algorithm. <br />The spectral sub-classes for each information class are then analyzed as a group and <br />subjected to appropriate merger, deletion, and lor augmentation. This is repeated for all sub- <br />classes of each information class. The entire collection of spectral sub-classes (for all <br /> <br />Page 19 <br />